Research On Fault Diagnosis Of EMU Rolling Bearing Based On VMD And ANFIS

Rolling bearing is an important part of high-speed EMU.It works in complex environment,faces various loads and different ways of wear.It is a component with high damage rate in rotating machinery.This paper takes the axle box bearing of running part of EMU as the research object,and studies the fault diagnosis method of rolling bearing of EMU in depth and compares the relevant diagnosis technology,The research and analysis,a set of fault diagnosis scheme of rolling bearing of EMU based on variable mode decomposition(VMD)and adaptive network based fuzzy information systems(ANFIS)is generated,which improves the accuracy and efficiency of fault diagnosis,so as to reduce the occurrence rate and prevent accidents.The work contents and results are as follows:(1)The background and present situation of fault diagnosis of rolling bearing are studied and summarized.The fault diagnosis mode and signal processing technology are summarized and sorted out,and the advantages and disadvantages of different algorithms and technologies are analyzed;The structure composition,damage mechanism,basic form of fault and characteristic frequency of vibration signal are summarized.(2)Wavelet packet signal decomposition is used to preprocess to reduce noise content.In the phase of noise reduction,this paper studies the wavelet packet threshold denoising,and obtains the best wavelet base function,decomposition layer number,threshold and threshold function through comparative analysis experiment.It solves the problem of over noise reduction and insufficient noise removal in the process of signal noise reduction,and can improve the accuracy of subsequent research.(3)The signal is decomposed by VMD and the eigenvalue is extracted to construct the feature vector.The empirical mode decomposition(EMD),set empirical mode decomposition(EEMD),local mean decomposition(LMD)and VMD decomposition methods are compared and simulated.Finally,the intrinsic mode function(IMF)components are calculated by VMD algorithm,The paper extracts different eigenvalues to construct the feature vector to input ANFIS system and analyzes the influence on the accuracy and stability of the system.Finally,the optimal feature vector construction is found to realize the application combination of VMD and ANFIS system which is most suitable for rolling bearing fault diagnosis.(4)The fault diagnosis and classification of rolling bearing is carried out by ANFIS algorithm.The system is based on BP neural network,and adopts fuzzy reasoning algorithm technology,which can get better learning ability and reasoning ability at the same time.(5)The system framework is built by computer,and the data from kassi university are simulated,tested and improved.Finally,the experiment is carried out.Through analyzing and improving the optimal system setting of parameter configuration,the validity and accuracy of the system are verified by using the real data of the laboratory.